Functional and pharmacological characterization of the modulatory role of serotonin on the firing activity of locus coeruleus norepinephrine neurons

Brain Research 922 (2001) 9–20 www.elsevier.com / locate / bres

Research report

Functional and pharmacological characterization of the modulatory role of serotonin on the firing activity of locus coeruleus norepinephrine neurons Steven T. Szabo a,b , Pierre Blier b , * b

a ´ , Canada H3 A 1 A1 Neurobiological Psychiatry Unit, McGill University, Montreal Department of Psychiatry, M c Knight Brain Institute, University of Florida, Gainesville, FL 32610 -0383, USA

Accepted 13 September 2001

Abstract Previous studies, using in vivo extracellular unitary recordings in anaesthetized rats, have shown that the selective 5-HT 1A receptor antagonist WAY 100,635 suppressed the firing rate of locus coeruleus (LC) norepinephrine (NE) neurons and that this effect was abolished by lesioning 5-HT neurons. In the present experiments, the selective 5-HT 2A receptor antagonist MDL 100,907, while having no effect on the spontaneous firing activity of LC neurons in controls, was able to restore NE neuronal discharges following the injection of WAY 100,635. The 5-HT 1A receptor agonist 8-OH-DPAT enhanced the firing activity of NE neurons and this action was entirely dependent on intact 5-HT neurons, unlike the inhibitory effect of the 5-HT 2 receptor agonist DOI. Taken together, these data indicate that 5-HT 2A but not 5-HT 1A receptors controlling LC firing activity are postsynaptic to 5-HT neurons. Prolonged, but not subacute, administration of selective 5-HT reuptake inhibitors (SSRIs) produces a decrease in the spontaneous firing activity of LC NE neurons. MDL 100,907 partially reversed this suppressed firing activity of LC neurons in paroxetine-treated rats. Although the a 2 -adrenoceptor antagonist idazoxan also enhanced the firing activity of NE neurons in paroxetine-treated rats, this increase was similar to that obtained in controls. In conclusion, prolonged SSRI treatment enhances a tonic inhibitory influence by 5-HT on LC neurons through postsynaptic 5-HT 2A receptors that are not located on NE neurons. A speculative neuronal circuitry accounting for these phenomena on LC NE activity is proposed.  2001 Published by Elsevier Science B.V. Theme: Neurotransmitters, modulators, transporters, and receptors Topic: Interactions between neurotransmitters Keywords: 5-HT 1A receptor; 5-HT 2A receptor; a 2 -Adrenoceptor; Antidepressant; Anxiety disorder; Firing activity

1. Introduction It is well established that norepinephrine (NE) neurons modulate the serotonin (5-HT) system. In the brainstem, dorsal raphe (DR) 5-HT neurons receive ascending NE neuron afferents originating from the locus coeruleus (LC), a nucleus almost exclusively responsible for the NE innervation of the forebrain [4,1,37,47]. Interactions between LC and DR impart a significant NE influence on the 5-HT system [47,32] and evidence has recently accumulated for the inverse relationship between these two nuclei *Corresponding author. Tel.: 11-352-392-3681; fax: 11-352-3922579. E-mail address: [email protected] (P. Blier).

as well, for review see Refs. [31] and [48]. For instance, Kaehler et al. [35] reported that descending projections from the DR to the LC account for at least 50% of the 5-HT innervation of this nucleus. These reciprocal monoaminergic interactions have been linked to the efficacy of antidepressant drugs in anxiety and affective disorders, in which treatment, but not necessarily etiology, of these disorders relies on altered NE and 5-HT transmission [15]. Interestingly, not all antidepressant drugs induce these neurochemical changes via the same mechanisms and agents that selectively target only one of these systems almost invariably produce alterations in both monoaminergic systems after chronic administration [9]. The selective serotonin reuptake inhibitors (SSRIs) are now considered as a first line therapeutic approach in the

0006-8993 / 01 / $ – see front matter  2001 Published by Elsevier Science B.V. PII: S0006-8993( 01 )03121-3

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

10

treatment of affective and many anxiety disorders. However, as with all antidepressants, they require an administration of at least 2 weeks before a clear beneficial effect can manifest itself. Long-term, but not acute or short-term SSRI administration increases the net output of 5-HT neurons, via the desensitization of 5-HT 1A and 5-HT 1B autoreceptors in the presence of 5-HT reuptake blockade, and induces an attenuation in the spontaneous firing activity of LC NE neurons [6,54,55] with a time-course that correlates with their anxiolytic and antidepressant responses (see [53] for review). Various 5-HT receptor subtypes, especially the 5-HT 1A and 5-HT 2A receptors, are believed to be implicated in the antidepressant and anti-panic effects of long-term SSRI treatment. Activation of these two receptor subtypes, which are present in the LC [39,57], alters NE neuron firing, however, in opposite directions. On the one hand, the enhancement of the firing rate of LC neurons produced by the intravenous administration of the 5-HT 1A receptor agonist 8-OH-DPAT is abolished in rats treated with antidepressants which selectively inhibit either 5-HT or NE transporter reuptake [50,51,55]. On the other hand, the attenuation of the firing rate of LC neurons induced by the 5-HT 2 receptor agonist DOI is decreased in long-term SSRI-treated rats, but unaltered by a NE reuptake blocker [55]. Thus, the therapeutic effect of drugs selective for the 5-HT system, like SSRIs, could be mediated in part by a modification of the efficacy of 5-HT transmission to the LC via the desensitization of 5-HT 1A and 5-HT 2 receptors, but this possibility needed to be better documented. In a first series of experiments, selective lesions of 5-HT neurons were performed to further delineate whether 5-HT neurons are necessary to mediate the 5-HT 1A and 5-HT 2A receptor effects that modulate NE neuron firing. In a second series of experiments, the effects of the selective 5-HT 2A and a 2 -adrenoceptor antagonists MDL 100,907 and idazoxan, respectively, were assessed on the firing activity of LC NE neurons in controls and 21-day paroxetine treated rats. The latter study was performed to ascertain whether enhanced 5-HT availability and 5-HT receptor activation / alterations in the LC from prolonged SSRI administration [33,54] may be mediating the attenuation of NE neuron activity by overactivating inhibitory 5-HT 2A receptors [52]. Given the reciprocal interactions between these two monoaminergic systems, it also had to be ensured that the attenuation on NE activity from sustained SSRI administration was truly a 5-HT mediated event and not to another receptor alteration. Thus, the response to idazoxan on LC firing activity was also evaluated in control and 21-day paroxetine treated rats.

2. Materials and methods

2.1. Animals The experiments were carried out in male Sprague–

´ Dawley rats (Charles River, St. Constant, Quebec, Canada) weighing between 300 and 325 g were kept under standard laboratory conditions (12:12 light–dark cycle with access to food and water ad libitum). Rats were anaesthetized with chloral hydrate (400 mg / kg, i.p.) and mounted in a stereotaxic apparatus (David Kopf Instuments). Supplemental doses (100 mg kg 21 , i.p.) were given to prevent any nociceptive reaction to pinching of the hind paw. Body temperature was maintained at 378C throughout the experiments utilizing a thermistor-controlled heating pad (Seabrook Medical Instruments). Prior to electrophysiological recording, rats were inserted with a catheter in a lateral tail vein for systemic injection of drugs.

2.2. Sustained treatment For the sustained treatment regimens, rats were anaesthetized with halothane containing a 2:1 O 2 –N 2 O mixture for subcutaneous implantation of osmotic minipumps (Alza, Palo Alto, CA, USA). Rats were treated with paroxetine (10 mg / kg / day) or saline for 21 days delivered by osmotic minipumps. These rats were tested with the minipumps in place in order to mimic the clinical condition whereby patients present an antidepressant response while taking their medication. Lesions of 5-HT neurons were performed on a separate group of rats that received, under chloral hydrate anesthesia (400 mg / kg, i.p.), intracerebroventricular (i.c.v.) injections of 5,7-dihydroxytryptamine (5,7-DHT; 200 mg of free base in 20 ml of 0.9% NaCl and 0.1% ascorbic acid) 1 h after the injection of desipramine (25 mg / kg, i.v.) to protect NE neurons from the neurotoxic action of 5,7-DHT. This protocol has previously been demonstrated to effectively decrease brain 5-HT levels [19]. A group of rats acting as controls for this experimental group underwent the same procedures but the solution injected i.c.v. only contained 0.9% saline. These rats were tested 10 days following the i.c.v. injection.

2.3. Electrophysiological experiments Extracellular unitary recordings of NE neurons were conducted with single-barrelled glass micropipettes preloaded with fiberglass filaments (to facilitate filling) being pulled in a conventional manner, with the tips broken back to 1–3 mm and filled with a 2 M NaCl solution. Their impedance range was between 2 and 4 MV. A burr hole was drilled 1 mm posterior to lambda and 1 mm lateral to midline for NE neurons recordings. Bleeding from disruption of the sagittal sinus was immediately stopped using bone wax. NE neurons were recorded with micropipettes lowered at 20.7 mm interaural and 1.1 to 1.4 mm lateral. Spontaneously active NE neurons of the LC were identified using the following criteria: regular firing rate (1–5 Hz) and positive action potential of long duration (0.8–1.2 ms) exhibiting a characteristic burst discharge in response to nociceptive pinch of the contralateral hind paw. Furthermore, by first locating the mesencephalic fifth motor

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

nucleus neurons that respond to lower jaw depression, moving medially to record LC NE neurons provided additional indication for the validity of the recordings. NE neurons were recorded for at least 1 min to establish basal firing rate. The selective 5-HT 2A and a 2 -adrenoceptor antagonists MDL 100,907 (200 mg / kg, i.v.) and idazoxan (1 mg / kg, i.v.), respectively, were injected in control and 21-day paroxetine treated rats while recording the spontaneous firing activity of the first NE neuron encountered before and after injection. Dose–response curves for the alteration of LC neuron firing activity were obtained for systemic (i.v.) administration of MDL 100,907 (200 mg / kg) after a preinjection of the selective 5-HT 1A receptor antagonist WAY 100635 (100 mg / kg) in untreated rats. In 5,7-DHT and control treated rats, dose–response effects to the preferential 5-HT 2A receptor agonist DOI [3,42,58,50] and the 5-HT 1A receptor agonist 8-OH-DPAT on NE neuron firing was assessed. Changes in the firing activity are expressed as percentages of baseline firing rate and were measured after systemic drug administration. Dose–response curves were constructed for only one dose of 8-OH-DPAT and DOI in each rat. However, in experiments where WAY 100,635 and MDL 100,907 were systemically administered, only one dose of MDL 100,907 preceded by the WAY 100,635 injection in each rat was used to generate an effective dose 50 (ED 50 ). Similarly, only one dose of DOI following an MDL 100,907 preinjection was used to generate an ED 50.

2.4. Drugs The following drugs were used: MDL 100,907 from Marion Merrell Dow (Cincinnati, OH, USA); 8-OH-DPAT, WAY 100,635, DOI, clonidine, ritanserin, desipramine and idazoxan from RBI (Natick, MA, USA). The concentrations and the doses used for these compounds were chosen on the basis of previous successful experiments carried out in our and other laboratories. Drugs administered i.v. were all dissolved in distilled water while ritanserin was dissolved dropwise by acetic acid then titrated with distilled water to the appropriate concentration.

11

2.5. Statistical comparisons All results were expressed as mean (6S.E.M.) of single neuron values. Statistical comparisons between the difference in NE neuron firing activity before and after intravenous drug administration were carried out using a paired t-test. Statistical comparisons of values obtained in treated and control rats were carried out using either one- or two-way analysis of variance. All posthoc pairwise multiple comparisons were done by the Student–Newman– Keuls method to assess the differences between controls and treated groups. Correlational coefficients (r values) for the dose–response relationship observed in the LC were calculated using simple linear regression analysis. The S.E.M. for the ED 50 values were calculated using regression analysis, with the y value of 50 used as the regressor. Difference between the two regressions were assessed by comparing their ED 50 values using the confidence intervals method. The Student’s t distribution was used to determine the 95% confidence limit, as well as for assessing the differences reported in mean NE neuron firing activity between the control and treated groups (Table 1).

3. Results

3.1. Effect of MDL 100,907 on WAY 100,635 -mediated attenuation of NE neuron firing activity Systemic injection of the selective 5-HT 1A receptor antagonist WAY 100,635 decreases the firing rate of NE neurons which can be prevented by a previous administration of drugs with 5-HT 2A antagonistic properties, but that were not selective for the latter receptor [31]. Given that MDL 100,907 is currently the only selective 5-HT 2A receptor antagonist available, it was deemed mandatory to assess whether this compound could reverse, rather than prevent, the effect of WAY 100,635 on LC firing activity. In the present study, acute injection of WAY 100,635 decreased the firing activity of NE neurons (|91%), as previously reported (Fig. 1A and B). This inhibitory effect of WAY 100,635 was reversed to the initial basal discharge values by a subsequent injection of the selective 5-HT 2A receptor antagonist MDL 100,907 (Fig. 1B), thus pro-

Table 1 Spontaneous firing activity (Hz) of LC NE neurons in controls and pre-treated rats Basil firing activity of NE neurons

Range of firing activity

Number of neurons †

Control 5,7-DHT

2.360.2 3.260.3*

0.9–3.7 2.1–5.6

22 10

Control Paroxetine

2.660.2 1.960.2*

1.6–3.9 1.1–2.8

12 10

Basal values refer to firing rates prior to any acute drug challenge. * P,0.05 when comparing treatment groups to controls using Student’s t-test. † Each value corresponds to the firing activity of single LC NE neurons recorded in each rat before drug injection. These values were obtained from all the data points in the dose-response curves.

12

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

Fig. 1. (A) Integrated firing rate histogram of a LC NE neuron illustrating the effect of intravenous administration of the selective 5-HT 1A antagonist WAY 100,635 in a control rat, producing a decrease in the firing activity, and that of a subsequent injection of the selective 5-HT 2A antagonist MDL 100,907 reversing it. (B) Mean effects of systemic administration of WAY 100,635 and then MDL 100,907 on the firing activity of LC NE neurons. *, P,0.001 (paired t-test) when compared to NE neuron basal firing activity of the before WAY 100,635 drug injections. †, P,0.001 (paired t-test) when compared to NE neuron basal firing activity before MDL 100,907 drug injections.

viding evidence of the latter receptor mediating these effects.

3.2. Effect of 8 -OH-DPAT and DOI on NE neuron firing activity in 5,7 -DHT treated rats The 5-HT 1A receptor agonist 8-OH-DPAT enhances the firing activity of LC NE neurons in a dose-dependent manner (Fig. 2A) [45,55]. The injection of the selective 5-HT 1A receptor antagonist WAY 100,635 shuts down firing activity following the injection of 8-OH-DPAT, an example of which is provided in Fig. 2A. This figure also highlights that a final injection of MDL 100,907 reversed

Fig. 2. (A) Integrated firing rate histogram of a LC NE neuron illustrating the effects of intravenous administration of the 5-HT 1A agonist 8-OHDPAT producing an increase in the firing activity, a subsequent injection of the selective 5-HT 1A antagonist WAY 100,635 attenuating the firing activity, and a final injection of the selective 5-HT 2A antagonist MDL 100,907 reversing the inhibitory action of WAY 100,635 in a control rat. (B) Integrated firing rate histogram of a LC NE neuron showing its absence of response to intravenous 8-OH-DPAT in a rat treated with the selective 5-HT neurotoxin 5,7-DHT. Note that the subsequent intravenous injection of the a 2 -adrenoceptor agonist clonidine was still able to attenuate NE neuron firing activity which was reversed by the a 2 adrenoceptor antagonist idazoxan.

the inhibitory effects of WAY 100,635 and returned NE firing activity back to basal values. The preferential 5HT 2A receptor agonist DOI suppresses the firing activity of NE neurons [17,55] and a following injection of the 5-HT 2A / 2C receptor antagonist ritanserin counteracted these effects (Fig. 3A). 5,7-DHT is a neurotoxin that destroys 5-HT neurons [19] as well as producing an increase in LC NE neuronal firing rate [31]. To assess whether the effects of the aforementioned 5-HT agonists on NE neuron firing are mediated via 5-HT neurons projecting to the LC, lesions with 5,7-DHT were performed. Consistent with previous findings [31], the mean spontaneous firing activity of NE neurons in 5,7-DHT

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

13

treated rats increased NE neuron firing to the same extent as that observed in control rats when idazoxan was the first drug injected (P50.844; Figs. 7A and 8). The response to injections of DOI in 5,7-DHT treated rats was blunted (ED 50 568 mg / kg) as compared to controls (ED 50 530 mg / kg; Fig. 4B). The dose–response curve for DOI in 5,7-DHT treated rats was thus shifted approximately 2-fold to the right (Fig. 4B). As in controls, an injection of ritanserin (n52) effectively reversed the inhibitory effects of DOI on NE neuron firing activity in 5,7-DHT treated rats (Fig. 3B).

3.3. Effect of MDL 100,907 on the firing activity of NE neurons in control and long-term paroxetine-treated rats

Fig. 3. (A) Integrated firing rate histogram of a LC NE neuron, recorded in a control rat illustrating the effects of intravenous administration of the preferential 5-HT 2A agonist DOI producing a decrease in the firing activity, and a subsequent injection of the 5-HT 2 antagonist ritanserin reversing the latter effect. (B) Integrated firing rate histogram of a LC NE neuron, recorded in a rat treated with the selective 5-HT neurotoxin 5,7-DHT illustrating the effect of intravenous administration of the preferential 5-HT 2A agonist DOI, and the subsequent injection of the 5-HT 2 antagonist ritanserin reversing the latter effect.

treated rats was increased as compared to controls (Table 1). Dose–response curves for systemic 8-OH-DPAT and DOI administration were obtained in control and 5,7-DHT treated rats (Fig. 4). The responsiveness of NE neurons to 8-OH-DPAT in 5,7-DHT treated rats was abolished over wide a range of doses, even at doses two times higher than the maximal excitatory dose obtained in the control group (Figs. 2A and 4A). To ascertain the integrity of the a 2 -adrenoceptors on NE neurons following the lack of 8-OH-DPAT response observed in 5,7-DHT treated rats, subsequent injections of the a 2 -adrenoceptor agonist clonidine suppressed the firing activity of NE neurons, being reversed by the a 2 -adrenoceptor antagonist idazoxan (n54; Fig. 2B). In fact, the injection of idazoxan (1 mg / kg, i.v.) following 8-OH-DPAT and clonidine in

Acute injections of the selective 5-HT 2A receptor antagonist MDL 100,907 did not alter the firing activity of NE neurons in control rats, but blocked the inhibitory effects of a subsequent injection of the preferential 5-HT 2A receptor agonist DOI (Figs. 4B and 6). In fact, MDL 100,907 was able to antagonize the suppressant effects of DOI on NE neuron firing activity at doses as high as 120 mg / kg of the agonist. After observing the lack of effect of DOI on NE neuron firing activity, clonidine was still able to decrease NE neuron firing activity while idazoxan reversed this effect (n52), an example of which is provided in Fig. 5A. The suppressant effects of long-term paroxetine treatment on the basal firing rate of the NE neurons was significant as compared to that of controls (Table 1; P50.01). The enhancing effect of MDL 100,907 (ca., 25%) on the firing activity of attenuated NE neurons in long-term paroxetine treated rats was statistically significant (P50.003), but this antagonist did not modify NE neuron basal firing rate at all in controls (Fig. 6). Interestingly, following the injection of MDL 100,907 in rats treated with paroxetine for 21-days, the injection of idazoxan further increased the firing activity of LC neurons (n52 rats; Fig. 5B).

3.4. Effect of idazoxan on the firing activity of NE neurons in control and long-term paroxetine treated rats The blockade of a 2 -adrenoceptors, which exert a negative feedback influence on NE neuron firing, has been shown to disinhibit these neurons resulting in an increase in firing rate in naive animals [26]. Indeed, the selective a 2 -adrenoceptor antagonist idazoxan increased the firing activity of NE neurons, an example of which is provided in Fig. 7A. The enhancing effect of idazoxan on NE neuron firing activity in control (|54%) and 21-day paroxetine treated rats (|62%) were significant (P50.004 and P5 0.007, respectively), but not statistically different from each other (P50.673; Fig. 8). Following the acute administration of idazoxan, which increased the attenuated NE neuron firing activity in 21-day paroxetine treated rats, a subsequent injection of MDL 100,907 (n53) was still

14

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

Fig. 4. (A) Relationship between the degree of augmentation of LC NE firing activity and doses of 8-OH-DPAT administered intravenously in controls and rats treated with the 5-HT neurotoxin 5,7-DHT. (B) Relationship between the degree of suppression of LC NE firing activity and doses of DOI administered intravenously in controls, 5,7-DHT treated rats, and rats receiving a preinjection of MDL 100,907. Only the initial response of a single NE neuron to the first dose of 8-OH-DPAT or DOI in each rat was used to construct the curves, with exception to rats that received a prior injection of MDL 100,907, DOI was the next drug injected. Outer lines represent the standard error of the regression lines. The shift to the right in the DOI dose–response curve in 5,7-DHT treated rats was significant as compared to the control. The 8-OH-DPAT and the DOI dose–response curves obtained in control rats were previously reported in [47], and used herein because these experimental series were performed concurrently.

able to enhance further the firing rate (Fig. 7B). When the basal firing rates of NE neurons in control and paroxetine treated rats that later received either MDL 100,907 or idazoxan, were merged within their respective groups, a statistically significant decrease was observed (Table 1).

4. Discussion The results of the present study first indicate the inhibitory effects of the selective 5-HT 1A receptor antago-

nist WAY 100,635 on LC NE neuron firing is mediated through an augmented 5-HT transmission at 5-HT 2A receptors. This is based on the premise that the suppressant effect of WAY 100,635 on NE neuron firing is dependent on intact 5-HT neurons, as their destruction prevents its inhibitory action [33]. Then, acute administration of the selective 5-HT 2A receptor antagonist MDL 100,907 by itself did not alter the firing activity of NE neurons (Fig. 6), but reversed the inhibitory effect of a prior injection of WAY 100,635 (Fig. 1). This finding is consistent with prior results whereby non-selective 5-HT 2 receptor antagonists,

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

Fig. 5. (A) Integrated firing rate histogram of a LC NE neuron illustrating the lack of effect of the intravenous administration of the selective 5-HT 2A receptor antagonist MDL 100,907 on firing activity and a subsequent injection the preferential 5-HT 2A agonist DOI (see Fig. 3, for an example of the inhibitory action of DOI). Intravenous injection of the a 2 -adrenoceptor agonist clonidine was still able to attenuate NE neuron firing activity and the a 2 -adrenoceptor antagonist idazoxan reversed this effect. (B) Integrated firing rate histogram of a LC NE neuron showing a increase in firing activity to intravenous MDL 100,907 in a rat treated with paroxetine (10 mg kg 21 day 21 ).

as well as the 5-HT 2A receptor antagonist spiperone, prevented the suppressant effect of WAY 100,635 on NE neuron firing [31]. Thus, administration of WAY 100,635 most likely increases 5-HT transmission [20,25] and decreases NE firing activity by an overactivation of excitatory 5-HT 2A receptors on inhibitory neurons projecting to the LC (Fig. 9) because neurons in this nucleus do not possess mRNA for 5-HT 2A receptors [44], and these receptors are not located in the raphe nuclei [24,18]. The spontaneous firing activity of LC NE neurons in 5,7-DHT treated rats was significantly increased as compared to that of their respective controls recorded in the present study (Table 1). This degree of enhancement is fully consistent with that previously reported [31], and therefore with the putative inhibitory role of 5-HT on the firing activity of these neurons. As expected, and contrary to results obtained with the 5-HT 1A antagonist WAY 100,635, activation of 5-HT 1A receptors using the 5-HT 1A agonist 8-OH-DPAT produced an excitatory effect on NE

15

Fig. 6. Effect of systemic administration of the selective 5-HT 2A receptor antagonist MDL 100,907 on the firing activity of LC NE neurons in control and paroxetine treated (10 mg kg 21 day) rats. †, P,0.05 (Student–Newman–Keuls method) when comparing LC basal firing activity in controls to paroxetine treated rats. *, P50.003 (paired t-test) when compared to NE neuron firing activity before MDL 100,907 drug injections. The number of neurons recorded is displayed in each box.

neuron firing activity which was also dependent on intact 5-HT neurons (Figs. 2 and 4A). This lack of effect of 8-OH-DPAT on LC firing activity in 5,7-DHT lesioned rats indicates that these receptors are not on LC neurons. Interestingly, these 5-HT 1A receptors desensitize after long-term administration with antidepressant drugs that selectively block either 5-HT or NE transporters [50,51,55]. Given that 5-HT neurons exert an inhibitory role on LC activity, and that 5-HT 1A autoreceptors located on the cell body of 5-HT neurons exert a negative feedback control on the firing activity of 5-HT neurons and ultimately the impulse-dependent release of 5-HT (Fig. 9; for review see [12]), these somatodendritic 5-HT 1A autoreceptors appear as candidates for mediating the effect of 8-OH-DPAT on NE neuron firing. In addition, it is widely accepted that this receptor population becomes subsensitive after long-term SSRI treatment and is important to produce a net increase in 5-HT transmission in projection structures during this time period [11,34,36,49]. However, there is considerable evidence indicating that the excitatory effects of 8-OH-DPAT on NE neuron firing is not mediated by somatodentritic 5-HT 1A receptors, but rather by the

16

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

Fig. 7. (A) Integrated firing rate histograms of LC NE neurons illustrating the excitatory effect of the intravenous administration of the selective a 2 -adrenoceptor antagonist idazoxan recorded in a control rat. (B) Note that the subsequent injection of the selective 5-HT 2A antagonist MDL 100,907 in the rat treated with paroxetine produced a further enhancement of firing.

activation of a different population of 5-HT 1A receptors. First, a dose of 5 mg / kg of 8-OH-DPAT completely shuts off DR 5-HT neuronal activity but leaves unaffected that of NE neurons (Fig. 4A). Second, acute administration of WAY 100,635 does not consistently alter 5-HT neuron firing activity by itself but shuts down that of NE neurons in the LC (Fig. 1A) [31]. Third, the selective NE reuptake blockers desipramine and reboxetine, which do not alter the sensitivity of somatodendritic 5-HT 1A autoreceptors [10,51], desensitize 5-HT 1A receptors controlling LC firing rate [55,51]. Nevertheless, due to the disappearance of the effect of 8-OH-DPAT and WAY 100,635 on LC firing in 5,7-DHT treated rats (Fig. 2B and [32]), it can be concluded that the latter action of these two 5-HT 1A receptor ligands depends on the integrity of 5-HT neurons. The location of the 5-HT 1A receptors, which control NE neuron firing, is probably not on the cell body of NE

Fig. 8. Effect of systemic administration of the a 2 -adrenoceptor antagonist idazoxan (1 mg kg 21 ) on the firing activity of LC NE neurons in control and paroxetine treated (10 mg kg 21 day) rats. *, P50.004 and P50.007 (paired t-test) when compared to the basal firing activity of the before idazoxan injection value in control and paroxetine treated rats, respectively. The number of neurons recorded is displayed in each box.

neurons, because neurons in the LC lack mRNA for 5-HT 1A receptors [45]. Considerable evidence has accumulated for a glutamate regulation of 5-HT release [41,5]. In fact, Van Bockstaele [56] recently used immunogold-silver labeling of an antibody that recognizes kainate receptors and peroxidase labeling for 5-HT neurons. These results showed that 5-HT terminals were labeled with kainate receptor immunoreactivity in the LC. Surprisingly however, 5-HT attenuates the activation of LC neurons by exogenous glutamate applications [16,2,14]. This dampened response of LC neurons was initially deemed to act through 5-HT 1A receptors on NE neurons as the iontophoretic application of 8-OH-DPAT produced the same effect [13]. Another study did not, however, confirm the latter observation [30], possibly due to methodological differences. Indeed, the use of high ejection currents in the former study certainly lead to a large outflow of chloride ions through the barrel used to counterbalance them. Such a large ionic flow may alter on its own neuronal responsiveness. The possibility that 8-OH-DPAT could alter the

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

17

Fig. 9. Speculative neuroanatomical and neurochemical bases for the interactions between 5-HT neurons and LC NE neurons. This diagram was prepared on the basis of the results obtained by several groups of investigators as well as data generated in our laboratory (see Discussion).

excitatory action of glutamate on NE neuronal firing appears somewhat puzzling given the absence of mRNA for 5-HT 1A receptors in the LC proper [43]. On the other hand, Weissman-Nanopoulos et al. [57] reported that following a 5,7-DHT lesion in rats, 5-HT 1A binding density in the LC was unchanged. Moreover, glutamate release from terminals in the LC is attenuated by perfusion of 8-OH-DPAT of LC slices [7]. Taken together, these data suggest the presence of inhibitory 5-HT 1A receptors on a glutamate neuron projecting to the LC. This interpretation is consistent with that proposed by Singewald and Phillippu [48] whereby the impact of 5-HT 1A ligands on LC activity is dependent on intact 5-HT neurons (Fig. 4A and [31]). One of the major afferent glutamatergic inputs to the LC is from neurons in the nucleus paragigantocellularis. In light of their importance in mediating sensory information to the LC as well as controlling NE neuron firing [21], 5-HT 1A receptors which impart an influence on LC firing may be located on neurons from the paragigantocellularis nucleus projecting to the LC. The activation of these 5-HT 1A receptors by systemic administration of 8-OHDPAT would then decrease the amount of glutamate available to depolarize 5-HT terminals, thereby inducing an attenuated activation of excitatory 5-HT 2A receptors on GABA, finally resulting in an increase in NE neuron firing (Figs. 2A and 9). Consistent with this proposed mechanism, it was recently reported that microiontophoretic application of the glutamate antagonist kynurenate and

bicuculline antagonize the modulatory effects of 8-OHDPAT and WAY 100,635, respectively, on LC NE neurons firing [52]. The suppressant effect of the preferential 5-HT 2A receptor agonist DOI on NE neuron firing activity was only decreased in 5,7-DHT treated rats as evidenced by a small but still significant shift to the right of its dose–response curve (Fig. 4B), when compared to the excitatory effect of 8-OH-DPAT which was abolished (Fig. 4A). Thus, 5-HT 2A receptors mediating the inhibitory effects of DOI on NE neuron firing activity can be considered as being predominantly postsynaptic to 5-HT neurons, contrary to the 5-HT 1A receptors which control LC firing that are entirely dependent on intact 5-HT neurons to exert their action. Also, it was previously shown that 5-HT depletion with the synthesis inhibitor p-chlorophenylalanine did not alter the inhibitory response of 5-HT 2 ligands on LC activity [29], but increased LC activity [46]. A high density of 5-HT 2A receptor labeling has been visualized in the rat and primate LC area using [ 3 H]MDL 100,907 [38,39]. These 5-HT 2A receptors are, however, probably not located directly on NE neurons because cell bodies labeled for the presence of the 5-HT 2A receptor protein were not detected within this nucleus, although the pericoerulear area was labeled and the nucleus prepositius hypoglossi as well [23,24]. The latter structure is another major afferent to the LC and exerts an inhibitory role on LC firing through GABAA receptors [22]. In addition to the anatomical evidence demonstrating a lack of 5-HT 2A receptors on LC neurons,

18

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

microiontophoretic application of DOI on LC neurons does not alter spontaneous or glutamate-induced NE neuron firing activity [30,13]. It has previously been reported that the inhibitory effects of intravenously injected DOI on LC firing can be blocked with the microiontophoretic application of the GABAA receptor antagonist bicuculline on LC neurons [17], and systemic injection of MDL 100,907 [50]. Furthermore, the inhibitory effect of DOI on LC neuronal firing is abolished upon hypoglossal nucleus destruction [30], therefore emphasizing the importance of this structure. However, DOI injected directly into the hypoglossal nucleus does not alter NE neuron activity [30]. Interestingly, 5-HT neurons synapsing on axon terminals making contact with tyrosine hydroxylase labeled neurons in the LC accounted for the greatest proportion of 5-HT elements in the LC [56]. It is thus proposed that, on the basis of these prior results and those of the present study, an excitatory 5-HT 2A receptor located on a hypoglossal GABA terminal projecting to the LC is primarily responsible for mediating the inhibitory effect of DOI on LC firing (Fig. 9). MDL 100,907 was able to reverse the inhibitory effects of WAY 100,635 on NE neurons firing activity presumably by counteracting the enhanced activation of such 5-HT 2A receptors (Fig. 9). This served as the impetus to assess whether an increase in 5-HT 2A receptor activation resulting from enhanced 5-HT transmission in long-term SSRItreated rats was contributing to the attenuation in LC firing activity observed after a 21-day treatment [54,55]. Consistent with the latter effect of SSRIs, Freuo et al. [27] observed that blood flow was attenuated to the greatest extent attenuated by the same treatment in this nucleus. Administration of MDL 100,907 in paroxetine-treated rats did produce an increase in NE neuron firing activity, as compared to controls where MDL 100,907 was totally inactive (Fig. 6). Even though the response to DOI in long-term SSRI treated rats is decreased [55], antagonism of these 5-HT 2A receptors could still enhance the firing activity of NE neurons, and perhaps not exactly to the control level precisely because of their partial desensitization (Fig. 6). In analogy, the increased availability of 5-HT in mice lacking 5-HT transporters results in a 3-fold decrease in 5-HT neuron firing activity, despite a marked desensitization of 5-HT 1A autoreceptors. Nevertheless, administration of WAY 100,635 reverses the attenuated firing activity in these mice [28]. It is important to mention that the decrease in LC firing activity observed in longterm paroxetine [54] and citalopram treated rats [55] is not likely due to increased a 2 -adrenoceptor activation because administration of the selective a 2 -adrenoceptor antagonist idazoxan produced the same increase in LC firing activity as compared to controls (Fig. 8). Furthermore, MDL 100,907 still increased the firing of LC neurons after the injection of idazoxan in paroxetine-treated rats (see Fig. 7B), thus supporting the possibility of enhanced 5-HT 2A receptor activation by long-term SSRI administration.

It appears that the desensitization of the 5-HT 1A receptors which control LC firing activity is common to all antidepressant drug classes tested thus far and may represent an important finding with respect to the treatment of certain anxiety disorders. Prolonged administration of SSRIs, which are effective in panic, generalized, and social anxiety disorders, desensitizes not only somatodendritic 5-HT 1A autoreceptors, but also 5-HT 1A receptors which normally augment LC firing, presumably through a decrease in GABA transmission. This is a possible mechanism that may explain, in part, the discrepancy between the rapid anxiolytic action of benzodiazepines, as a result of their facilitatory action on the GABAA receptor, and the delayed response obtained with SSRIs. Consistent with this possibility are reports of decreased GABA levels and / or GABAA / benzodiazepine receptor binding in several brain areas of depressive and panic disorder patients [40,8]. In conclusion, the present study unveiled complex interactions between 5-HT and NE neurons. These also appear to involve other neurotransmitters, such as GABA and glutamate through which the modulatory effect of SSRIs on spontaneous LC neuronal firing would be exerted. Further experiments should help establish the precise localization of the various receptors involved in this complex circuitry.

Acknowledgements This work was supported by a Medical Research Government of Canada (MRC) Grant (MT-11410 to P.B.), a MRC doctoral award (MRC-1554 to STS), a Fonds pour la Formation de Chercheurs et L’Aide a` la recherche-Fonds ´ ´ de la Recherche en Sante´ du Quebec (FRSQ-FCAR-Sante68803 to STS), and by salary support from the University of Florida to PB and STS.

References [1] C. Anderson, D.A. Pasquez, W.B. Forbes, P.J. Morgane, Locus coeruleus-to-dorsal raphe input examined by electrophysiological and morphological methods, Brain Res. Mol. Brain Res. 23 (1994) 163–178. [2] G. Aston-Jones, H. Akaoka, P.J. Charlety, G. Chouvet, Serotonin selectively attenuates glutamate-evoked activation of noradrenergic locus coeruleus neurons, J. Neurosci. 11 (1991) 760–769. [3] C.S. Aulakh, P. Mazzola-Pomietto, B.A. Hulihan-Giblin, D.L. Murphy, Lack of cross-tolerance for hypophagia induced by DOI versus m-CPP suggests separate mediation by 5-HT 2A and 5-HT 2C receptors, respectively, Neuropsychopharmacology 13 (1995) 1–8. [4] J.M. Baraban, G.K. Aghajanian, Suppression of serotonergic neuronal firing by a-adrenoceptor antagonists: evidence against GABA mediation, Eur. J. Pharmacol. 66 (1980) 287–294. ´ In vivo evidence for an inhibitory [5] D. Becquet, M. Faudon, F. Hery, glutamatergic control of serotonin release in the cat caudate nucleus: involvement of GABA neurons, Brain Res. 519 (1990) 82–88.

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20 ´ [6] J.C. Beıque, C. de Montigny, P. Blier, G. Debonnel, Blockade of 5-HT and NE reuptake by venlafaxine: a comparative study with paroxetine and desipramine, Br. J. Pharmacol. 125 (1998) 526–532. [7] D.H. Bobker, J.T. Williams, Serotonin agonists inhibit synaptic potentials in the rat locus ceruleus in vitro via 5-hydroxytryptamine1A and 5-hydroxytryptamine1B receptors, J. Pharmacol. Exp. Ther. 250 (1989) 37–43. [8] J.D. Bremner, R.B. Innis, T. White, M. Fujita, D. Silbersweig, A.W. Goddard, L. Staib, E. Stern, A. Cappiello, S. Woods, R. Bladwin, D.S. Charney, SPECT [I-123]iomazenil measurement of the benzodiazepine receptor in panic disorder, Biol. Psychiat. 47 (2000) 96–106. [9] P. Blier, N. Haddjeri, S.T. Szabo, J. Dong, Enhancement of serotonin neurotransmission — a sometimes insufficient course of antidepressant action, Hum. Psychophamacol. Clin. Exp. 16 (2001) 23–27. [10] P. Blier, C. de Montigny, Effect of chronic tricyclic antidepressant treatment on the serotonergic autoreceptor: a microiontophoeretic study in the rat, Naunyn-Schmiedeberg’s Arch. Pharmacol. 314 (1980) 123–128. [11] P. Blier, C. de Montigny, Electrophysiological investigations on the effect of repeated zimelidine administration on serotonergic neurotransmission in the rat, J. Neurosci. 3 (1983) 1270–1278. [12] P. Blier, C. de Montigny, Serotonergic neurons and 5-HT receptors in the CNS, Handb. Exp. Pharmacol. 129 (1997) 727–750. [13] P.J. Charlety, G. Aston-Jones, H. Akaoka, M. Buda, G. Chouvet, Participation of 5-HT 1A receptors in the decrease by serotonin of activation of locus coeruleus neurons by glutamate, Comp. Rend. l’Acad. Sci. III. 312 (1991) 421–426. [14] P.J. Charlety, K. Chergui, H. Akaoka, C.F. Saunier, M. Buda, G. Aston-Jones, G. Chouvet, Serotonin differentially modulates responses mediated by specific excitatory amino acid receptors in the rat locus coeruleus, Eur. J. Neurosci. 5 (1993) 1024–1028. [15] D.S. Charney, G.R. Heninger, Abnormal regulation of noradrenergic function in panic disorders. Effects of clonidine in healthy subjects and patients with agoraphobia and panic disorder, Arch. Gen. Psychiat. 43 (1986) 1042–1054. [16] G. Chouvet, H. Akaoka, G. Aston-Jones, Selective reduction by serotonin of neuronal excitation in the locus coeruleus by glutamate, Compt. Rend. l’Acad. Sci. III 306 (1988) 339–344. [17] C. Chiang, G. Aston-Jones, A 5-hydroxytryptamine 2 agonist augments g-aminobutyric acid and excitatory amino acid inputs to noradrenergic locus coeruleus neurons, Neuroscience 54 (1993) 417–420. [18] V. Cornea-Hebert, M. Riad, C. Wu, S.K. Singh, L. Descarries, Cellular and subcellular distribution of the serotonin 5-HT 2A receptor in the central nervous system of adult rat, J. Comp. Neurol. 409 (1999) 187–209. [19] C. de Montigny, R.Y. Wang, T.A. Reader, G.K. Aghajanian, Monoaminergic denervation of the rat hippocampus: microiontophoretic studies on pre- and postsynaptic supersensitivity to norepinephrine and serotonin, Brain Res. 200 (1980) 363–376. [20] L.J. Dreshfield-Ahmad, D.C. Thompson, J.M. Schaus, D.T. Wong, Enhancement in extracellular serotonin levels by 5-hydroxytryptophan loading on administration of WAY 100635 and fluoxetine, Life Sci. 66 (2000) 2035–2041. [21] M. Ennis, G. Aston-Jones, Activation of locus coeruleus from nucleus paragigantocellularis: a new excitatory amino acid pathway in brain, J. Neurosci. 10 (1988) 3644–3657. [22] M. Ennis, G. Aston-Jones, GABA-mediated inhibition of locus coeruleus from the dorsomedial rostral medulla, J. Neurosci. 9 (1989) 2973–2981. [23] R. Fay, L. Kubin, Pontomedullary distribution of 5-HT 2A receptorlike protein in the rat, J. Comp. Neurol. 418 (2000) 323–345. [24] C.T. Fischette, B. Nock, K. Renner, Effects of 5,7-dihydroxytryptamine on serotonin1 and serotonin2 receptors throughout the rat central nervous system using quantitative autoradiography, Brain Res. 421 (1987) 263–279.

19

[25] C.A. Fornal, F.J. Martin, A. Mendlin, C.W. Meltzer, B. Bjorvatn, B.L. Jacobs, Pindolol increases extracellular 5-HT while inhibiting serotonergic neuronal activity, Eur. J. Pharmacol. 377 (1999) 187– 191. [26] J.E. Freedman, G.K. Aghajanian, Idazoxan (RX 781094) selectively antagonizes a 2-adrenoceptors on rat central neurons, Eur. J. Pharmacol. 105 (1984) 265–272. [27] U. Freo, C. Ori, M. Dam, A. Merico, G. Pizzolato, Effects of acute and chronic treatment with fluoxetine on regional glucose cerebral metabolism in rats: implications for clinical therapies, Brain Res. 854 (2000) 35–41. [28] G. Gobbi, D.L. Murphy, K.P. Lesch, P. Blier, Modifications of the serotonergic system in mice lacking serotonin transporters: an in vivo electrophysiological study, J. Pharmacol. Exp. Ther. 296 (2001) 1–9. [29] E. Gorea, J. Adrien, Serotonergic regulation of noradrenergic coerulean neurons: electrophysiologic for the involvement of 5-HT 2 receptors, Eur. J. Pharmacol. 154 (1988) 285–291. [30] E. Gorea, D. Davenne, L. Lanfumey, M. Chastanet, J. Adrien, Regulation of noradrenergic coerulean neuronal firing mediated by 5-HT 2 receptors: involvement of the prepositus hypoglossal nucleus, Neuropharmacology 30 (1991) 1309–1318. [31] N. Haddjeri, C. de Montigny, P. Blier, Modulation of the firing activity of noradrenergic neurones in the rat locus coeruleus by the 5-hydroxytryptamine system, Br. J. Pharmacol. 120 (1997) 865– 875. [32] H. Hebert, C.B. Saper, Organization of medullary adrenergic and noradrenergic projections to the periaqueductal gray matter in the rat, J. Comp. Neurol. 315 (1992) 34–52. [33] P.D. Hrdina, B. Foy, A. Hepner, R.J. Summers, Antidepressant binding sites in brain: autoradiographic comparison of [ 3 H]paroxetine and [ 3 H]imipramine localization and relationship to serotonin transporter, J. Pharmacol. Exp. Ther. 252 (1990) 10–18. [34] T. Jolas, S. Haj-Dahmane, E.J. Kidd, X. Langlois, L. Lanfumey, C.M. Fattaccini, V. Vantalon, A.M. Laporte, J. Adrien, H. Gozlan, Central pre- and postsynaptic 5-HT 1A receptors in rats treated chronically with a novel antidepressant, cericlamine, J. Pharmacol. Exp. Ther. 268 (1994) 1432–1443. [35] S.T. Kaehler, N. Singewald, A. Philippu, Dependence of serotonin release in the locus coeruleus on dorsal raphe neuron activity, Naunyn Schmiedeberg’s Arch. Pharmacol. 359 (1999) 386–393. [36] D.S. Kreiss, I. Lucki, Effects of acute and repeated administration of antidepressant drugs on extracellular levels of 5-hydroxytryptamine measured in vivo, J. Pharmacol. Exp. Ther. 274 (1995) 866–876. [37] L.A. Loizou, Projections of the nucleus locus coeruleus in the albino rat, Brain Res. 15 (1969) 563–569. [38] J.F. Lopez-Gimenez, G. Mengod, J.M. Palacios, M.T. Vilaro, Selective visualization of rat brain 5-HT 2A receptors by autoradiography with MDL 100,907, Naunyn-Schmiedeberg’s Arch. Pharmacol. 356 (1997) 446–454. [39] J.F. Lopez-Gimenez, M.T. Vilaro, J.M. Palacios, G. Mengod, [ 3 H]MDL 100,907 labels 5-HT 2A serotonin receptors selectively in primate brain, Neuropharmacology 37 (1998) 1147–1158. [40] A.L. Malizia, V.L. Cunningham, C.J. Bell, P.F. Liddle, T. Jones, D.J. Nutt, Decreased brain GABA(A)-benzodiazepine receptor binding in panic disorder: preliminary results from a quantitative PET study, Arch. Gen. Psychiat. 55 (1998) 715–720. [41] M.L. Mayer, G.L. Westbrook, The physiology of excitatory amino acids in the vertebrate central nervous system, Prog. Neurobiol. 28 (1997) 197–267. [42] P. Mazzola-Pomietto, C.S. Aulakh, K.M. Wozniak, J.L. Hill, D.L. Murphy, Evidence that 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane (DOI)-induced hyperthermia in rats is mediated by stimulation of 5-HT 2A receptors, Psychopharmacology 117 (1995) 193– 199. [43] M. Pompeiano, J.M. Palacious, G. Mengod, Distribution and cellular localization of mRNA coding for 5-HT 1A receptor in the rat brain: correlation with receptor binding, J. Neurosci. 12 (1992) 440–453.

20

S.T. Szabo, P. Blier / Brain Research 922 (2001) 9 – 20

[44] M. Pompeiano, J.M. Palacios, G. Mengod, Distribution of the serotonin 5-HT2 receptor family mRNAs: comparison between 5HT2A and 5-HT2C receptors, Brain Res. Mol. Brain Res. 23 (1994) 163–178. [45] M.F. Piercey, M.W. Smith, J.T. Lum-Ragan, Excitation of noradrenergic cell firing by 5-hydroxytryptamine 1A agonist correlates with dopamine antagonist properties, J. Pharmacol. Exp. Ther. 268 (1993) 1297–1303. [46] T.A. Reader, R. Briere, L. Grondin, A. Ferron, Effects of pchlorophenylalanine on cortical monoamines and on the activity of noradrenergic neurons, Neurochem. Res. 11 (1986) 1025–1035. [47] K. Sakai, D. Salvert, M. Touret, M. Jouvet, Afferent connections of the nucleus raphe dorsalis in the cat as visualized by the horseradish peroxidase technique, Brain Res. 137 (1977) 11–35. [48] N. Singewald, A. Philippu, Release of neurotransmitters in the locus coeruleus, Prog. Neurobiol. 56 (1998) 236–267. [49] J. Sprouse, J. Brazelton, L. Renolds, 5-HT 1A agonist potential of pindolol: electrophysiological studies in the dorsal raphe nucleus and hippocampus, Biol. Psychiat. 47 (2000) 1050–1055. [50] S.T. Szabo, P. Blier, SSRI plus 5-HT 2A receptor antagonism compared to 5-HT reuptake inhibition alone produces inverse effects on the firing activity of noradrenergic neurons, Int. J. Neuropsychopharmacol. Suppl. 1 (2000) S214. [51] S.T. Szabo, P. Blier, Effect of acute and sustained administration of reboxetine on locus coeruleus noradrenergic and dorsal raphe 5-HT neurons, Eur. J. Neurosci. 13 (2001) 2077–2087.

[52] S.T. Szabo, P. Blier, Serotonin 1A receptor ligands act on norepinephrine neuron firing through excitatory amino acid and GABAA receptors: A microiontophoretic study in the rat locus coeruleus, Synapse 42 (2001) 203–212. [53] S.T. Szabo, P. Blier, Response of the norepinephrine system to antidepressant drugs, CNS Spectrums 6 (2001) 679–684. [54] S.T. Szabo, C. de Montigny, P. Blier, Modulation of noradrenergic neuronal firing by selective serotonin reuptake blockers, Br. J. Pharmacol. 126 (1999) 568–571. [55] S.T. Szabo, C. de Montigny, P. Blier, Progressive attenuation of the firing activity of locus coeruleus noradrenergic neurons by sustained administration of selective serotonin reuptake inhibitors, Int. J. Neuropsychopharmacol. 3 (2000) 1–11. [56] E.J. Van Bockstaele, Multiple substrates for serotonergic modulation of rat locus coeruleus neurons: relationships with kainate receptors, Brain Res. Bull. 51 (2000) 433–442. [57] D. Weissmann-Nanopoulos, E. Mach, Y. Magre, J. Demassey, J.F. Pujol, Evidence for the localization of 5-HT 1A binding sites on serotonin containing neurons in the raphe dorsalis and raphe centralis nuclei of the rat brain, Neurochem. Int. 7 (1985) 1061– 1073. [58] J. Yamada, Y. Sugimoto, K. Horisaka, Serotonin 2 (5-HT2) receptor agonist 1-(2,5-dimethoxy-4-iodophenyl)-2-aminopropane (DOI) inhibits chlorpromazine and haloperidol-induced hypothermia in mice, Biol. Pharmacol. Bull. 18 (1995) 1580–1583.